Secretagogue-dependent Phosphorylation of the Insulin Granule Membrane Protein Phogrin Is Mediated by cAMP-dependent Protein Kinase*

Phogrin, a 60/64-kDa integral membrane protein of dense-core granules in neuroendocrine cells, is phosphorylated in a Ca 2 (cid:1) -sensitive manner in response to secretagogue stimulation of pancreatic (cid:2) -cells. Phosphorylation of the phogrin cytosolic domain by (cid:2) -cell homogenates was Ca 2 (cid:1) -independent but stimulated by cAMP. Recombinant protein kinase A (PKA) could phosphorylate phogrin directly. High performance liquid chromatography analysis of tryptic phosphopeptides, combined with site-directed mutagenesis of candidate sites, revealed the presence of two phosphorylation sites at Ser-680 and Thr-699, located in the juxtamembrane region between the transmembrane span and the pro-tein-tyrosine phosphatase homology domain of phogrin. Full-length wild-type phogrin, as well as mutant ver-sions where Ser-680 and Thr-699 had been replaced either by alanines or by aspartic acid residues, were targeted to secretory granules in transfected AtT20 neuroendocrine cells. Stimulation of these cells with a range of secretagogues, including K (cid:1) , BaCl 2 , and forskolin, demonstrated that the in vivo phosphorylation sites are the same as those identified in vitro . In MIN6 (cid:2) -cells, the PKA inhibitor H-89 prevented Ca 2 (cid:1) -dependent phogrin

Protein phosphorylation is thought to play a central role in the regulation of hormone release from dense-core secretory granules (1)(2)(3). Stimulation of pancreatic ␤-cells with glucose leads to depolarization of the plasma membrane, Ca 2ϩ influx via voltage-gated channels, and an increase in free cytosolic Ca 2ϩ (4,5). Ca 2ϩ regulates early as well as late events in granule exocytosis (5), possibly via the activation of Ca 2ϩ -dependent protein kinases, including Ca 2ϩ /calmodulin-dependent kinase II (CaMKII) 1 (6), myosin light chain kinase (7), and Ca 2ϩ -dependent isoforms of protein kinase C (8). However, unequivocal evidence for functional involvement of these enzymes in the secretory process is still lacking.
Other second messengers, in particular cAMP, contribute to the regulation of exocytosis (9 -11). Insulinotropic hormones such as glucagon-like peptide-1 or pituitary adenylyl cyclaseactivating polypeptide increase intracellular cAMP levels by stimulating adenylyl cyclase through G-protein coupled receptors on the cell surface, an effect that can be mimicked by drugs like forskolin, which activate adenylyl cyclase directly. A rise in cAMP, and the subsequent activation of protein kinase A (PKA; cAMP-dependent kinase), is generally not sufficient in itself to evoke a secretory response in ␤-cells, but significantly potentiates glucose-induced insulin release. A number of mechanisms appear to be stimulated by cAMP (reviewed in Ref. 10), including membrane depolarization via the ATP-dependent K ϩ channel (12) and non-selective cation channels (13), an increase in Ca 2ϩ influx through voltage-gated L-type Ca 2ϩ channels (14 -17), enhanced Ca 2ϩ mobilization from intracellular stores (18), and the recruitment of secretory granules from a reserve pool to the release sites (19,20).
Despite numerous studies implicating secretagogue-dependent activation of protein kinases in regulated secretion, the underlying molecular mechanisms and cellular targets of phosphorylation are poorly characterized. A large number of ␤-cell proteins are phosphorylated in a Ca 2ϩ -or cAMP-dependent manner in cell-free extracts (reviewed in Ref. 3), but for the most part they have not been identified. In intact ␤-cells, the analysis of known in vitro kinase substrates as candidate proteins has demonstrated stimulus-dependent phosphorylation of synapsin I (21,22) and microtubule-associated protein MAP-2 (23) by CaMKII, and of the glucose transporter GLUT-2 by PKA (24). It is not clear, however, if these phosphorylation events play a direct role in mediating insulin release.
Phogrin (phosphatase homologue in granules of insulinoma) is a transmembrane glycoprotein localized to dense-core secretory granules in a wide range of neuronal and endocrine cell types (25,26). It is a member of the protein-tyrosine phosphatase (PTP) family, with a single cytosolic PTP domain, but appears to lack tyrosine phosphatase activity due to the substitution of a highly conserved residue within the substrate binding pocket. The protein is initially synthesized as a 135-kDa precursor and post-translationally processed in the secretory pathway to the mature 60/64-kDa form. We have recently shown that phogrin is phosphorylated in intact MIN6 ␤-cells upon secretagogue stimulation (27). The time course of phosphorylation parallels that of insulin release, and, like secretion, phosphorylation in response to glucose or high K ϩ is dependent on the presence of Ca 2ϩ . Conditions that increase cAMP levels stimulate phogrin phosphorylation in a Ca 2ϩ -independent manner. The modification of phogrin by two different intracellular signaling systems, both important in regulated exocytosis, combined with its strategic location on the dense-core granule surface, make it a good candidate for a protein with a key role in the regulation of secretory granule function.
In this study, we report that phogrin is a substrate for PKA, and that phosphorylation occurs on a Ser and a Thr residue in its cytosolic domain. Our results indicate that Ca 2ϩ -and cAMP-mediated responses target the same phosphorylation sites, with the two signaling pathways converging upstream of PKA activation.

MATERIALS AND METHODS
Reagents-All chemicals were purchased from Sigma or Fisher, unless indicated otherwise. Molecular biology reagents were from New England Biolabs; cell culture supplies were obtained from Life Technologies, Inc.; forskolin, IBMX, KN-93, and H-89 were from Calbiochem.
In Vitro Phosphorylation of Phogrin: Analysis of Ca 2ϩ Dependence-Glutathione S-transferase (GST) and the cytosolic domain of rat phogrin fused to GST (GST/C) (25) were purified by glutathione affinity chromatography. GST or GST/C (5 M) was incubated at 30°C with MIN6 extract (2 g of protein) in 50 l of 40 mM HEPES, pH 7.0, 10 mM MgCl 2 , 20 M ATP, 2 Ci of [␥-32 P]ATP (PerkinElmer Life Sciences) in the presence of 2 mM EGTA, 0.5 mM Ca 2ϩ , or 10 g/ml calmodulin as indicated in the figure legends. MIN6 extract was prepared by rinsing cells in phosphate-buffered saline (PBS) and homogenizing (several passes through a 28-gauge needle) in 40 mM HEPES, pH 7.4, 2 mM EGTA with protease inhibitors (10 M pepstatin A, 10 M E-64, 1 mM phenylmethylsulfonyl fluoride). The homogenate was centrifuged at 12,000 ϫ g for 10 min, the pellet extracted once in 0.25 volume of the above buffer containing 0.5% Triton X-100, re-centrifuged, and the supernatants combined. Cytosol fractionated by gel filtration (see below) was analyzed for phogrin phosphorylating activity in the same way, except that 25 l of each column fraction was used/50-l assay, resulting in a final concentration of 75 mM NaCl, 0.5 mM EGTA, and 1 mM CaCl 2 where present. Reactions were initiated by the addition of enzyme preparation, and terminated by adding trichloroacetic acid to a final concentration of 10% (w/v). Proteins were precipitated on ice for 60 min, pelleted by centrifugation at 14,000 ϫ g for 10 min, and washed once in cold acetone before resuspension in SDS-PAGE sample buffer and polyacrylamide gel electrophoresis. Gels were fixed in 50% (v/v) methanol, 10% (v/v) acetic acid, in some cases in the presence of Coomassie Blue to stain proteins, and dried. Phosphate incorporation was visualized by phosphorimaging (Storm 860; Molecular Dynamics) and quantitated using ImageQuant software (Molecular Dynamics).
CaMKII Assays-CaMKII activity was measured using the peptide substrate autocamtide-2. Peptide (25 M) was incubated with MIN6 cytosol fractions (25 l/50-l assay) as described above for GST/C. Reactions were terminated by addition of trichloroacetic acid to a final concentration of 5% (w/v), incubated on ice for 20 min, and cleared of precipitated protein by centrifuging for 1 min at 12,000 ϫ g. Aliquots of each supernatant were spotted onto P-81 paper (Whatman), immersed in cold 0.75% phosphoric acid, and rinsed three times in fresh solution. Incorporated radioactivity was detected by Cerenkov counting of filter papers in water.
Protein Kinase A-dependent in Vitro Phosphorylation-GST or GST/C (5 M) was incubated at 30°C in 50 l of 50 mM Tris, pH 7.4, 10 mM MgCl 2 , 50 M ATP, 2 Ci of [␥-32 P]ATP with MIN6 protein in the presence or absence of 10 M 8-Br-cAMP. MIN6 cells were either disrupted by sonication in PBS with protease inhibitors to obtain total protein (10 g/assay) or fractionated by gel filtration (5 l sample/ assay). (Note: this resulted in a dilution of NaCl carried over from column fractions to 15 mM and avoided inhibition of PKA activity by high NaCl.) Alternatively, recombinant PKA catalytic subunit (0.1-10 units) (New England Biolabs) was used. Reaction products were analyzed by SDS-PAGE as detailed above.
HPLC Analysis of Peptides-100 g of GST/C was phosphorylated with 5 units of PKA in the presence of [␥-32 P]ATP (8 Ci) in 100 l of 50 mM Tris, pH 7.4, 10 mM MgCl 2 , 50 M ATP for 2 h at 30°C. Samples were heated to 100°C for 15 min, cooled, and incubated with sequencing-grade trypsin (2 g of trypsin:100 g of GST/C) for 5 h at 37°C. Digestion was stopped by adding 5 l of 10% (v/v) trifluoroacetic acid, and insoluble material was removed by centrifuging for 5 min at 9,000 ϫ g. For HPLC fractionation (BioCAD 700E), peptides were loaded onto a C18 reverse phase column equilibrated with 0.1% trifluoroacetic acid and 10% (v/v) acetonitrile. Fractions of 1 ml, or 0.2 ml (in the peak 1 and 2 regions), were collected at a flow rate of 1 ml/min, using a linear gradient of 10 -60% (v/v) acetonitrile for the first 60 ml, then 60 -95% acetonitrile for 20 ml. Peptides were monitored by absorbance at 214 nm; 32 P was measured by Cerenkov counting of fractions. Peptides of interest were sequenced by Edman degradation at the UCHSC Cancer Center core facility.
Transfections-Phogrin constructs in pcDNA3 were transfected into AtT20 cells with LipofectAMINE (Life Technologies, Inc.), following the manufacturer's recommendations. Stable neomycin-resistant cell lines were selected with 1 mg/ml G418 and screened for phogrin expression by Western blotting and immunofluorescence microscopy. For each construct, several phogrin-positive cell lines were initially characterized, and representative lines were chosen for subsequent experiments. Cells were maintained in medium containing 0.25 mg/ml G418.
Western Blotting-Cells were rinsed in PBS and scraped into PBS containing protease inhibitors for the determination of protein concentrations (Bio-Rad Protein Assay). Alternatively, cells were lysed directly in SDS-PAGE sample buffer containing dithiothreitol. Gel electrophoresis, semidry transfer to nitrocellulose membrane, and immunodetection by enhanced chemifluorescence were performed as described previously (27). Phogrin was detected using an affinity-purified antibody raised to the lumenal domain of the protein (25). Antisera to PKA (catalytic subunit) were from StressGen (anti-C␣; 1:2000), and from Santa Cruz Biotechnology (anti-C␣/␤; 1:500).
Bandshift Assays-MIN6 cells were pre-incubated for 2 h under basal conditions (modified Krebs buffer without glucose), then stimulated for 40 min with 16.7 mM glucose or 10 M forskolin in the same buffer. Kinase inhibitors were added 30 min before the end of the pre-incubation period. Cells were harvested in SDS-PAGE sample buffer, subjected to electrophoresis on 7% gels to separate the 60-and 64-kDa forms of phogrin, and analyzed by Western blotting.

Phosphorylation of Recombinant Phogrin in Vitro-
We have previously shown that glucose-stimulated phogrin phosphorylation in intact ␤-cells is dependent on the presence of extracellular Ca 2ϩ . Inhibitor studies targeting different protein kinases were consistent with CaMKII activation being involved (27). To examine this phosphorylation event further, the phogrin cytoplasmic domain fused to GST (GST/C) was incubated in vitro with MIN6 ␤-cell protein extract in the presence of [␥-32 P]ATP. Analysis on SDS-PAGE gels showed phosphate incorporation into the fusion protein (Fig. 1A). However, this was not stimulated by the addition of Ca 2ϩ and/or calmodulin and still occurred in the presence of a Ca 2ϩ chelator (Fig. 1B). Fractionation of MIN6 cytosol by gel filtration revealed that CaMKII activity eluted as a high molecular weight peak (Fig.  1C, top panel), consistent with the existence of the enzyme in vivo as a multimeric complex of 8 -12 subunits (6). Ca 2ϩ /calmodulin also stimulated the incorporation of phosphate into several endogenous proteins in the same fractions (the bands of about 55 kDa on SDS-PAGE gels may represent autophosphorylated CaMKII monomers); however, no phosphorylation of phogrin was seen under these conditions (bottom panels). Similarly, phogrin was a poor substrate for recombinant CaMKII (data not shown).
In vitro phosphorylation of phogrin GST/C by MIN6 extract was markedly stimulated by cAMP or its non-hydrolyzable analogue 8-Br-cAMP ( Fig. 2A). In MIN6 cytosol fractionated by gel filtration, cAMP-stimulated phosphorylating activity eluted in two broad peaks (Fig. 2B, upper panels). Western blot analysis of PKA distribution with antibodies to the catalytic subunit of the kinase showed a similar profile, both for the ␣and the ␤-isoforms (lower panels). The enzyme's presence in high molecular weight fractions may reflect the formation of a complex with regulatory subunits, and presumably interactions with anchoring proteins (28). Incubation of GST/C with the recombinant catalytic subunit of PKA alone phosphorylated the fusion protein directly, in a concentration-dependent manner (Fig. 2C).
Phosphorylation of mature phogrin in pancreatic ␤-cells induces a characteristic mobility shift on SDS-PAGE gels, giving rise to a 60-kDa basal form and a phosphorylated protein migrating with an apparent molecular mass of 64 kDa. When in vitro phosphorylated GST/C was examined on SDS-PAGE gels, three distinct molecular forms were observed after Coomassie staining. Phosphorimaging revealed that two of these had incorporated phosphate (Fig. 2D). Phosphorylated form 1 (P1) migrated with an apparent molecular mass of around 2 kDa higher than non-phosphorylated GST/C; P2 showed a shift of about 4 kDa. In a time-course experiment, P1 was observed initially, and the P2 form at later time points (data not shown), indicating the existence of more than one phosphorylation site.
Identification of PKA Phosphorylation Sites-The cytosolic domain of phogrin contains 56 serine or threonine residues, 7 of these in classic PKA recognition motifs (29). To narrow down the location of the target sites, phogrin GST/C phosphorylated in vitro by recombinant PKA in the presence of [␥-32 P]ATP was subjected to proteolytic digestion with trypsin and fractionated by HPLC. Two peaks of 32 P incorporation were detected (peaks 1 and 2) (Fig. 3A), in both cases corresponding to well resolved peptide peaks monitored as absorbance at 214 nm (Fig. 3B). Comparison of profiles for tryptic peptides derived from phosphorylated and non-phosphorylated GST/C demonstrated a shift of the two A 214 peaks toward elution at lower solvent  5). B, upper panels, cAMP-stimulated phogrin phosphorylating activity was determined in cytosol fractionated by gel filtration. GST/C was incubated with fractions (5 l/assay) for 20 min as in A. Lower panels, fractions (40 l) were subjected to SDS-PAGE, transferred to nitrocellulose membrane, and Western-blotted for PKA using antibodies recognizing either the ␣-isoform or the ␣and ␤-isoforms of the catalytic subunit. C, GST/C was incubated with recombinant PKA catalytic subunit and [␥-32 P]ATP. Left panel, incubation with 1 unit of PKA for 1-30 min; right panel, incubation for 30 min with 0.1-10 units of enzyme. (Note: left panel shows a longer exposure than right panel.) D, GST/C was incubated as above with or without PKA. Gels were run for an extended period of time to separate closely spaced bands. Proteins were stained with Coomassie Blue, followed by phosphorimaging to determine incorporation of radioactivity.
concentrations upon incorporation of negatively charged phosphate groups into the peptides.
N-terminal sequencing of peptide 1 yielded Ile-Asn-Ser-Val-Ser, identifying it as a 19-residue tryptic peptide spanning amino acids 678 -696. For peptide 2, Ser-Ser-Thr-Ser-Ser was obtained, corresponding to a 32-residue peptide from amino acids 697 to 728 (Fig. 3C). Both peptides form part of a region highly enriched in serines that is located between the trans-membrane span and the PTP homology domain of phogrin.
The majority of PKA sites conform to the consensus targets Arg-X-(Ser/Thr) or Arg-X-X-Ser/Thr (29). Based on this, sitedirected mutagenesis was performed on candidate residues within peptides 1 and 2. Mutants were expressed as cytosolic domain GST fusion proteins and tested for in vitro phosphorylation by recombinant PKA. Substitution of Ser-680 with alanine resulted in a significant decrease in phosphate incorporation, and in a reduced mobility shift on SDS-PAGE gels (Fig.  4A). Even after prolonged incubation times, S680A did not reach the P2 position of fully phosphorylated GST/C. Additional point mutations introduced at either Ser-698 or Ser-700 did not significantly alter the phosphorylation pattern. However, a S680A/T699A double mutant neither incorporated phosphate nor showed the corresponding bandshift (Fig. 4B), suggesting that Ser-680 and Thr-699 were the PKA target sites in rat phogrin.
Subcellular Localization and Phosphorylation of Recombinant Phogrin in AtT20 Cells-AtT20 pituitary corticotroph tumor cells were chosen to examine the behavior of mutant forms FIG. 3. HPLC analysis of tryptic phosphopeptides. A, phogrin GST/C was phosphorylated with PKA catalytic subunit in the presence of [␥-32 P]ATP and subjected to digestion with trypsin. Peptides were separated by reverse phase HPLC using a gradient of acetonitrile for elution. Fractions of 1 ml were collected, and incorporated radioactivity (cpm) was determined. B, HPLC fractionation was repeated exactly as in A, except that 0.2-ml fractions were collected in the peak 1 and 2 regions. Incorporation of phosphate (cpm) is shown in the top panels, peptide profiles (A 214 ) in the middle panels (P-GST/C), and the corresponding peptide profiles derived from non-phosphorylated GST/C in the bottom panels. Arrows indicate the peptide peaks undergoing a phosphorylation-induced shift. C, N-terminal amino acid sequences obtained from the peak 1 and peak 2 peptides, respectively, are shown in bold. The full sequence of each tryptic peptide is given below, with arrows denoting the trypsin cleavage sites. The diagram indicates the positions of peptides 1 and 2 within mature 60/64-kDa phogrin. of phogrin in a cellular context because these cells express very low levels of the endogenous molecule (Fig. 5, A (lane 1) and B (first panel)). Cells stably transfected with full-length wild-type phogrin analyzed by immunoblotting with an antibody directed against the lumenal domain of the protein showed a doublet band at 60/64 kDa, the expected size for mature, post-translationally processed phogrin (Fig. 5A). In addition, significant levels of the 135-kDa precursor and a 71/75-kDa doublet, presumably a processing intermediate, were seen.
Immunofluorescence microscopy did not detect significant phogrin expression in the parental cell line or in cells transfected with vector alone. Cells transfected with wild-type phogrin showed strong labeling at the tips of cell extensions, as well as in the perinuclear region and in punctate structures throughout the cell body (Fig. 5B). The distribution of phogrin overlapped extensively with the labeling for the granule marker ACTH, both in the cell processes and in phogrin-positive cytoplasmic vesicles, suggesting that AtT20 cells, like pancreatic ␤-cells, targeted the recombinant molecule to secretory granules. When protein synthesis was inhibited with cycloheximide for 2.5 h, the perinuclear pool largely disappeared while labeling of the tips was unaffected. A low level of cell surface staining was also seen for phogrin, and a small number of phogrin-positive, ACTH-negative vesicles were evident, especially in the cytoplasm. This is consistent with differences in trafficking between granule membrane components, which are retained and are thought to recycle after exocytosis, and content proteins, which are released.
To evaluate the AtT20 cells' ability to phosphorylate phogrin in response to secretagogue stimulation, cells transfected with wild-type phogrin were pre-equilibrated with [ 32 P]P i , followed by incubation either in basal or in stimulation medium containing 10 M forskolin and 0.5 mM IBMX. Immunoprecipita-tion of phogrin showed little background in cells transfected with vector alone. In phogrin-expressing cells, some 32 P labeling was observed under basal conditions, which was markedly increased in response to forskolin stimulation (Fig. 5C). Phosphate was incorporated both into the fully processed 60/64-kDa phogrin and the 71/75-kDa intermediate form.
Expression of Phosphorylation Mutants in AtT20 Cells-To study phogrin phosphorylation in a cellular context, point mutations were introduced into the full-length molecule. The PKA target residues were replaced with either alanine to prevent their phosphorylation, or aspartic acid to mimic constitutively phosphorylated residues. Phogrin S680A/T699A (AA) or S680D/T699D (DD) appeared to be post-translationally processed as efficiently and reached steady-state levels similar to those for the wild-type molecule (Fig. 6A). The subcellular localization of phogrin/AA and phogrin/DD also resembled that of the wild-type protein, with prominent immunolabeling at the tips of cell processes and in the perinuclear region, as well as in punctate structures throughout the cell (Fig. 6B). For all three phogrin variants, the steady-state distribution showed extensive co-localization with the secretory granule marker ACTH. As observed for the wild-type molecule, inhibition of protein synthesis resulted in reduction of the perinuclear pool of phogrin/AA or DD, but did not affect staining in the tips of processes where the majority of granules are located (Fig. 6C).
To determine if phogrin phosphorylation in intact cells occurred at Ser-680 and Thr-699, and to examine if further residues in addition to the target sites identified in vitro were involved, cells expressing equivalent amounts (see Fig. 6A) of wild-type, AA or DD phogrin were analyzed by [ 32 P]P i labeling. Different types of secretagogues were used, including those acting via cAMP-dependent pathways (the adenylyl cyclase activator forskolin and the phosphodiesterase inhibitor IBMX), FIG. 5. Expression, localization, and phosphorylation of exogenous phogrin in AtT20 cells. A, proteins (40 g/ lane) from non-transfected cells or cells stably expressing full-length wild-type phogrin were separated by SDS-PAGE and Western-blotted for phogrin. i., intermediate; m., mature protein. B, cells transfected with either vector alone or phogrin were fixed for immunofluorescence, permeabilized, and double-labeled with polyclonal anti-phogrin and monoclonal anti-ACTH antibodies, followed by visualization with fluorescently tagged secondary antibodies. Phogrin is shown in green (Cy2), and ACTH in red (Cy3). Bottom panels show cells treated with 100 g/ml cycloheximide (CHX) for 2.5 h before fixation. Arrowheads mark the tips of cell processes; arrows indicate labeling in the perinuclear region. C, AtT20 cells stably transfected either with vector alone or with phogrin were pre-equilibrated with [ 32 P]P i in basal buffer for 2 h, followed by a 30-min incubation under basal or stimulating (10 M forskolin ϩ 0.5 mM IBMX) conditions. Phogrin was immunoprecipitated from cell lysates and subjected to SDS-PAGE and phosphorimaging. Arrows mark 60/64-kDa mature phogrin (m.) and the 71/75-kDa partially processed form (i.) and those relying on Ca 2ϩ -dependent pathways (depolarizing concentrations of K ϩ , or Ba 2ϩ , which mimics the effect of Ca 2ϩ on exocytosis). Stimulation of cells with either 55 mM KCl, 1 mM BaCl 2 or 10 M forskolin ϩ 0.5 mM IBMX, or with a combination of the above, resulted in the incorporation of phosphate into wild-type phogrin, but neither into the AA nor the DD mutants (Fig. 7). These findings suggested that Ser-680 and Thr-699 represent the physiologically relevant target sites for secretagogue-dependent phosphorylation.
PKA-dependent Versus Ca 2ϩ -dependent Phosphorylation-The highest levels of phogrin phosphorylation were achieved by treating cells with 10 M forskolin in the presence of 0.5 mM IBMX, a combination of compounds aimed at maximally stimulating PKA by increasing the rate of cAMP synthesis while inhibiting its breakdown by phosphodiesterases. Depolarization with K ϩ or stimulation with BaCl 2 resulted in lower but reproducible levels of phosphorylation. When the cells received a high K ϩ or a BaCl 2 stimulus in addition to forskolin/IBMX, phosphate incorporation into phogrin did not increase beyond that induced by forskolin/IBMX alone (Fig. 7). This is suggestive of the convergence of the Ca 2ϩ -dependent and the cAMP-  8B). Identical results were obtained with K ϩ -stimulated cells (data not shown). Taken together, this suggested that phogrin phosphorylation induced by Ca 2ϩ -dependent secretagogues required PKA activity, and that the targets of W-7 and KN-93 were upstream of PKA activation. DISCUSSION Phogrin phosphorylation in intact MIN6 ␤-cells in response to glucose or K ϩ only occurs in the presence of extracellular Ca 2ϩ , and is blocked by inhibitors of CaMKII. Ca 2ϩ -independent phosphorylation, however, can be evoked by cAMP, presumably via the activation of PKA (27). Attempts to reconstitute Ca 2ϩ -dependent phosphorylation in vitro using MIN6 homogenate, fractionated cytosol, or recombinant CaMKII as a source of enzyme were unsuccessful, although the kinase was clearly active under the assay conditions used. On the other hand, MIN6 preparations phosphorylated phogrin in vitro in a cAMP-dependent manner, an activity that co-fractionated with endogenous PKA and could be mimicked by recombinant PKA catalytic subunit alone.
Recombinant phogrin cytosolic domain, upon phosphorylation with PKA, underwent a mobility shift similar to that characteristic of phosphorylation of the native protein, supporting the hypothesis that the in vitro modifications reflect events in a stimulated ␤-cell. We showed that phosphorylation occurred at Ser-680 and Thr-699 on two adjacent tryptic peptides, both located within a stretch of around 50 amino acids containing a total of 18 serines and threonines. This cluster of serine residues is positioned between the transmembrane span and the PTP domain, a part of the molecule that is highly susceptible to proteases in native phogrin (25). Thus, this region may represent a flexible hinge linking the transmembrane segment to the more tightly folded globular PTP domain. In such a model, phosphorylation of residues in the hinge region could result in a conformational change within the cytoplasmic domain to permit or exclude interactions with binding partners.
To study the behavior of mutant phogrin molecules in a neuroendocrine cell type, AtT20 cells were chosen as a model system because of their low levels of endogenous protein. Exogenous phogrin was efficiently targeted to secretory granules when expressed at levels comparable to those found in MIN6 cells. Under steady-state conditions, the protein is present both in granules, which in AtT20 cells are concentrated in the tips of  Upper panel (lanes 1-3), cells were pre-equilibrated with [ 32 P]P i before stimulation, followed by immunoprecipitation of phogrin. Lower panel (lanes 1-5), phogrin was analyzed by Western blotting. Arrowheads indicate the phosphorylated (P) and non-phosphorylated (N) forms of the protein. B, cells were stimulated for 40 min with glucose (16.7 mM) in the presence or absence of forskolin and kinase inhibitors as shown. Phogrin phosphorylation was analyzed by Western blotting .   FIG. 7. Phosphorylation of phogrin/AA and phogrin/DD in AtT20 cells. Cells stably expressing wild-type phogrin, or either AA or DD mutant phogrin, were pre-equilibrated with [ 32 P]P i for 2 h, followed by a 30-min incubation in basal or stimulating buffer. Stimulation buffers contained 55 mM KCl (K ϩ ), 1 mM BaCl 2 (Ba 2ϩ ), 10 M forskolin ϩ 0.5 mM IBMX (F/I), or either KCl or BaCl 2 in combination with forskolin ϩ IBMX. As a control, cells transfected with vector alone were stimulated with forskolin ϩ IBMX. Phogrin was immunoprecipitated using an antiserum to the lumenal domain. The bottom panel represents quantitation of phosphate incorporation into the 60/64-kDa form of phogrin. Results shown are representative of two independent experiments. i., intermediate; m., mature protein; wt, wild-type.